How to Master Estimation
TLDR: Master Estimation by anchoring to everyday reference objects, rounding numbers into manageable factors, and building intuition about scale. There are three question pools · Comparisons, Products and Volumes · and you play exactly one of them at a time. Each asks for a different thing, and only one of the three gives you a list to choose from. One wrong answer ends your streak, so reason first - guess last.
What Is Estimation and Why It Matters
Estimation is a number-sense training game. Each round poses a question that you answer by judgement rather than by exact arithmetic - how many small things it takes to match one big thing, roughly what two numbers multiply to, or which of three drawn solids holds the most. Two of the three pools give you a list to pick from; the third asks you to type a number. One wrong answer ends your streak.
The game trains Fermi estimation: the art of making educated guesses using reasoning and reference points rather than precise arithmetic. Scientists, engineers, and analysts use this skill every day to sanity-check data, plan projects, and make decisions under uncertainty. Here you build the same intuition through three distinct question pools: Comparisons, Products, and Volumes.
One thing to know before you plan a session: the difficulty does not move. The game has no easy, medium or hard, and the ladder does not raise the setting as your streak grows · it holds one value for the whole run. The single thing a longer streak changes is the number of options in Comparisons, and that is finished by round three. A long run is a test of consistency, not of rising difficulty.
The Three Question Pools
Comparisons ask for a ratio between two everyday things, not for which one is larger. “How many popcorn kernels weigh the same as an elephant?” You pick the closest of four to six numbers. The six question shapes are weigh the same as, are needed to lift, lined up span, stacked up reach, fit inside, and cover the area of. The pool holds about 7,800 facts, and it leans long: roughly half are lengths, and the rest are weights, volumes and, rarely, areas.
Products ask for one multiplication, and this is the pool with no answer list. You get two numbers between 30 and about 110, and you type your estimate into a box. You are not graded on an exact match · you win when you land within 13 percent of the true product, and the reveal tells you both the exact answer and the band you had.
Volumes draw three different solids side by side at one shared scale · a box, cube, cylinder, sphere, cone, pyramid, triangular prism or tetrahedron · and ask which holds the most. You answer A, B or C. There is no container to fill and no fill level to read.
You cannot mix the pools. The game plays one pool at a time, and you choose which one in the Settings panel · or by adding ?topic=products or ?topic=volumes to the page address. Each pool keeps its own personal best, so switching never costs you a record.
Core Tactics for Strong Estimation
Anchor to Reference Objects
The foundation of Fermi estimation is anchoring - comparing unknowns to things you know well. Keep a mental catalogue of everyday scales:
- A human stands roughly 1.7 m tall and weighs about 70 kg
- Earth’s population is about 8 billion
- A standard room is roughly 4 m by 6 m
- A car is about 5 m long
- A grain of sand is roughly 0.5 mm across
- A litre of water weighs 1 kg
When asked to estimate any quantity, immediately ask: “What familiar object or number can I compare this to?” That anchor gives you a starting point you can reason from rather than guess at.
Tip: After each session, note one ratio that surprised you. (“I did not realise it takes tens of millions of popcorn kernels to weigh as much as one elephant.”) A surprising ratio sticks better than an expected one, and each one expands your reference library.
Round and Factor for Products
When Products shows you 68 x 94, decompose into friendly factors. Round 68 to 70 and 94 to 90, multiply 7 by 9 to get 63, then put the two zeros back: about 6,300. The true answer is 6,392, so that five-second estimate lands 1.4 percent out and wins comfortably inside the 13 percent band.
Round to the nearest five, not the nearest ten. Both factors sit between 30 and 109, and that range decides the rule. Round each one to the nearest five and multiply, and you land inside the 13 percent band on every pair the game can deal · all 6,400 of them, worst case 12.5 percent. Round to the nearest ten instead and you lose about one round in fifteen, because five is a sixth of a factor of thirty and two such errors can compound past 30 percent. The nearest five costs you no extra time and it never fails.
Visualise Volume Scaling
This pool is kinder than it looks, and it is worth knowing why. The winning solid is grown until it has both the largest volume and the largest outline on screen, so a sphere never hides extra volume behind a taller cylinder. The biggest-looking solid is the answer. Use the scaling rules below to confirm a close call, not to overturn what you see:
- Doubling the height of a cylinder doubles its volume (one dimension changes)
- Doubling the radius of a cylinder multiplies volume by four (quadratic scaling)
- Doubling all three dimensions of a box multiplies volume by eight (cubic scaling)
A solid twice as wide, twice as deep and twice as tall holds eight times more, not two. When two solids look close, ask which dimensions differ and how many of them are involved. The reveal prints each solid’s own formula and result, so a round you got wrong tells you exactly where your eye was fooled.
Tip: The gap between the winner and the others never narrows · the winner is always at least about a third larger than either rival, at every point in a run. If a Volumes round ever feels like a coin toss, you are reading the drawing wrong rather than meeting a harder round.
Narrow by Elimination
This works in Comparisons and Volumes, the two pools that give you a list. Products has no list, so skip ahead. Use elimination systematically:
- Identify the most extreme choices (smallest and largest). If one is clearly absurd, remove it.
- Reason about the middle ground. Does your estimate fall closer to the lower or upper cluster?
- Pick the answer that aligns with your reasoning, even without full certainty.
Logic beats luck. Narrowing four options to two already cuts your error rate in half. In Comparisons the wrong options are built as magnitude neighbours of the right one, so they are spaced by powers rather than scattered · once you fix the order of magnitude, usually one option is left.
Common Mistakes to Avoid
Losing Track of Magnitude: A single rounding step can shift an answer from millions to billions. After every mental calculation, confirm the scale: “Is this thousands, millions, or billions?” Scale errors are the single most common reason for a broken streak.
Overconfidence on Comparisons. These questions feel intuitive and hide traps, because a ratio of two familiar things is rarely a familiar number. It takes tens of millions of popcorn kernels to match one elephant, and the gap between “millions” and “tens of millions” is exactly where the wrong options live. Work out both sides in the same unit before you commit.
Forgetting that packing wastes space. When a Comparisons round asks how many small things fit inside a big one, the answer is not the plain volume ratio. The game assumes about 64 percent packing, the figure for loose spheres, so the true count is roughly two thirds of the number you get by dividing. The reveal shows that step.
Trusting gut instead of the reveal. Every round, right or wrong, ends with the full arithmetic: the division for a Comparison, the exact product and your error band for a Product, each solid’s formula for a Volume. A player who taps straight through that panel learns nothing from a run. Read it on the rounds you got right too.
Reverse-Check: After estimating, reason backward from your answer. If you chose 15 billion for a product, can you sketch why that magnitude is plausible? This takes three seconds and catches scale errors before you tap.
Difficulty Progression and Streak Strategy
There is no difficulty setting, and the streak does not raise one. The run holds a single difficulty from the first round to the last, so the Products numbers stay in the same 30-to-110 range, the 13 percent band never tightens, and the Volumes gap never narrows. The one thing that does move is the Comparisons option count: four options in round 1, five in round 2, and six from round 3 onward for the rest of the run.
Rounds 1 to 3: The only rounds that differ from the rest, and only in Comparisons, where a fourth then a fifth wrong option appears. Use them to warm up the habit: say why, out loud or in your head, even when the answer looks obvious.
Rounds 4 onward: Every round from here is the same shape as the last. This is where anchoring and rounding habits pay off, because nothing else is going to help you. Slow down; a two-second pause is cheaper than a wrong answer.
A long run: The enemy is not difficulty, it is attention. A streak ends because the thirtieth round got the same three seconds as the third, and one order of magnitude slipped. Treat a long streak as a test of holding the same routine, not of rising to a harder one.
Track Your Weak Pool. Each pool keeps its own personal best, so the three numbers already tell you where you are weakest · read them rather than guessing. Then open Settings, switch to that pool, and stay there for two or three sessions. Because you can only play one pool at a time anyway, this is not a sacrifice; it is just choosing the one that is worth the session.
A Practice Routine to Build Mastery
Session 1: Build your reference library (10 minutes). Play Comparisons, the pool the game opens on. After every answer - right or wrong - read the reveal and note what anchor you should have used. You are not chasing a streak; you are building mental scaffolding.
Session 2: Products only (10 minutes). Switch the pool to Products in Settings. State your rounding step, aloud or in your head, before you type. Then compare your number with the exact product in the reveal and notice which way you drifted.
Session 3: Volumes only (10 minutes). Switch the pool to Volumes. Before choosing, name how many dimensions differ and whether that makes the scaling linear, quadratic or cubic. Then check your reasoning against the formulas in the reveal.
Session 4: Your weakest pool (15 minutes). Go back to whichever of the three holds your lowest best and aim for one long streak in it. When it breaks at round 5, restart and aim to pass round 5. Gradual improvement beats marathon sessions.
Tip: Space sessions across separate days rather than cramming them into one afternoon. Number sense consolidates during sleep, and you will feel noticeably sharper the next morning than you did at the end of yesterday’s session.
Streak Success: Long streaks come from consistent reasoning, not lucky guesses. Anchor every estimate to something familiar, round big numbers into manageable factors, and reason backward to check the magnitude. Those three habits compound - each week your personal best will climb.
Beyond the Game: Real-World Estimation
The skills Estimation builds transfer directly. You will mentally check project budgets, travel times, and data sizes. You will catch unreasonable claims because you instinctively know rough scales. The discipline of anchoring and rounding is the same whether you are estimating the weight of a shipment or the user count of a new feature rollout.
Start with Comparisons to build your reference library, then take the other two pools in turn. Each wrong answer reveals a gap in your number sense, and the reveal names it for you. Each long streak confirms your intuition is sharpening. Keep practising - estimation becomes second nature faster than you expect.
Estimation
Test your number sense across three lenses: wild comparisons, product estimation, and volume guessing. All in one streak
Play nowWorks on any device.